Three-dimensional model visualization comparison system for construction acceptance

Abstract

The application discloses a three-dimensional model visualization comparison system for construction acceptance, and relates to the technical field of digital twinning and three-dimensional model simulation of construction engineering.The system is based on computer-aided design and digital twinning architecture, and comprises a measured data analysis and fusion module, a three-dimensional space deviation simulation calculation module, a dynamic color spectrum mapping and voxel warning module, and a viewpoint adaptive lead routing and label rendering module.The application has the advantage that the dynamic deviation label rendering technology with a lead is introduced in the three-dimensional scene, thereby breaking through the technical bottleneck that traditional comparison systems can mostly only output two-dimensional reports, text statistical charts or static pictures.The user can locate the specific out-of-tolerance position and the spatial form of the out-of-tolerance through in-situ and three-dimensional error display in the three-dimensional model simulation, thereby greatly improving the decision-making efficiency and accuracy in the construction site acceptance process.

Description

Technical Field

[0001] This invention relates to the field of digital twin and 3D model simulation technology for building engineering, specifically a 3D model visualization and comparison system for construction acceptance. Background Technology

[0002] With the rapid development of Building Information Modeling (BIM) and digital twin technologies, digital transformation in the engineering field has become an irreversible trend. In the construction and acceptance phase of building projects, accurately comparing the actual construction results (usually in the form of 3D point clouds, oblique photography, etc.) with the initial design BIM model is a core link in ensuring project quality and assessing construction errors. However, traditional construction acceptance relies heavily on manual on-site measurement, or although digital technology has been introduced, it only outputs the comparison results as two-dimensional charts, reports, or static two-dimensional textures.

[0003] In existing technical solutions, for example, Chinese patent document CN119026198B (authorization announcement number, Method and System for Comparing BIM Models and Drawings Based on As-Built Archives) discloses a system that uses natural language processing to parse as-built archives and uses a capsule network algorithm to compare BIM models with construction drawings. This prior art can automatically identify and annotate areas of difference, thereby generating a comparison analysis report. However, this prior art has obvious drawbacks: The final output of the comparative analysis results is a comparative analysis report that mainly includes error statistics, two-dimensional difference markers and textual risk assessment. It lacks a visualization mechanism that can truly and intuitively display error attributes in complex three-dimensional space. In actual large or complex structural engineering projects (such as irregular steel structures and factories with complex electromechanical pipelines), two-dimensional reports or simple location markers are difficult to intuitively display the three-dimensional distribution pattern of spatial deviations. When faced with hundreds or thousands of deviation markers, acceptance personnel often find it difficult to immediately identify the severity of the deviation and its specific three-dimensional spatial location due to occlusion of the three-dimensional view and the complexity of the model depth. This greatly reduces the efficiency and support of on-site visualization decision-making.

[0004] In addition to the aforementioned literature, there are also some label display methods based on general 3D rendering engines in the existing technology. However, these methods generally use fixed screen space text boxes to label model deviations. When users roam, zoom, or rotate their view in the 3D scene, the fixed text boxes are very easy to overlap, stack, or be occluded by the foreground and background of the model itself (i.e., occlusion culling fails). These conventional rendering methods do not take into account the spatial topological relationship of the digital twin model under a specific observation viewpoint, and lack viewpoint adaptive rendering mechanisms and deviation level warning linkage, making the 3D spatial positioning of deviation information virtually useless in actual operation.

[0005] In summary, developing a system that can render dynamic deviation labels and chromatogram out-of-tolerance warnings with leads in real time in a 3D scene, solves the problem that traditional reports cannot intuitively display the spatial deviation distribution and out-of-tolerance location, and avoids information occlusion and stacking in complex 3D interactive roaming, and achieves accurate 3D positioning of deviation information and clear visual decision support, has become a major technical challenge that urgently needs to be solved in the current field of this technology. Summary of the Invention

[0006] To overcome the technical shortcomings of existing technologies, such as low level of 3D visualization, easy overlap and occlusion of multi-dimensional deviation information, and difficulty in intuitively supporting spatial decision-making, this invention proposes a 3D model visualization comparison system for construction acceptance. This invention achieves accurate calculation, 3D positioning, and real-time anti-occlusion presentation of deviation information in complex construction systems by constructing a dynamic deviation label rendering mechanism based on spatial topology and viewpoint adaptation, and a chromatogram out-of-tolerance warning system based on point-to-surface spatial distance assessment.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A 3D model visualization comparison system for construction acceptance, the system is based on computer-aided design and digital twin architecture, including a measured data analysis and fusion module, a 3D spatial deviation simulation calculation module, a dynamic chromatographic mapping and voxel warning module, and a viewpoint adaptive routing and label rendering module; The measured data parsing and fusion module is used to acquire the three-dimensional reconstruction point cloud data of the construction site and the initial design three-dimensional model data. It uses a three-dimensional registration algorithm to register the three-dimensional reconstruction point cloud data to the unified global coordinate system where the initial design three-dimensional model data is located, so as to realize the fusion and alignment of multi-source heterogeneous models in three-dimensional space. The three-dimensional spatial deviation simulation calculation module is communicatively connected to the measured data parsing and fusion module. It is used to calculate the minimum Euclidean distance of each aligned measured data point relative to the surface of the initial design three-dimensional model using the nearest neighbor search space algorithm, extract all spatial deviation data, and establish a deviation dataset containing spatial coordinates and deviation amounts. The dynamic chromatographic mapping and voxel warning module is communicatively connected to the three-dimensional spatial deviation simulation calculation module. It is used to map the deviation amount in the deviation dataset into RGB rendering color values ​​in real time based on the preset out-of-tolerance threshold and chromatographic mapping function, and seamlessly attach and cover the RGB rendering color values ​​to the surface of the measured data points to form a three-dimensional chromatogram. Simultaneously, the global three-dimensional space is divided into multiple specific voxel blocks, and the out-of-tolerance density value within each specific voxel block is counted. When the out-of-tolerance density value exceeds the safety threshold, a pulsed highlight warning is triggered in the local space. The viewpoint adaptive routing and label rendering module is communicatively connected to the dynamic color mapping and voxel warning module, and is used to capture the observation angle of the current viewport camera in real time during digital twin 3D scene roaming. For anchor points with significant deviations in the deviation dataset, the optimal guide path and label floating coordinates that are non-overlapping and avoid obstruction by the model body are calculated using force-directed layout and view frustum anti-occlusion target optimization algorithm. Dynamic deviation labels with guides are then rendered in real time in the 3D scene.

[0008] Furthermore, in the dynamic chromatographic mapping and voxel warning module, for each measured data point, its distance function from the model surface and the corresponding chromatographic warning mapping mechanism depend on a preset mapping curve to ensure the prominence of out-of-tolerance parts at different observation levels.

[0009] Furthermore, in the viewpoint adaptive lead path routing and label rendering module, the dynamic generation of the lead path is achieved by constructing a viewpoint occlusion cost function and a lead path optimization objective function, ensuring that during the user's interaction of rotating and scaling the model, the deviation labels automatically undergo mechanical repulsion and position iterative updates, completely eliminating the intersection and overlap of labels in screen space.

[0010] Compared with the prior art, the beneficial effects of the present invention by adopting the above technical solution are as follows: (1) The present invention introduces dynamic deviation label rendering technology with lead wires in a three-dimensional scene, which breaks through the technical bottleneck that most traditional comparison systems can only output two-dimensional reports, text statistical charts or static pictures. By displaying the error in situ and three-dimensionally in the three-dimensional model simulation, users can clearly locate the specific out-of-tolerance position and the spatial form of the out-of-tolerance, which greatly improves the decision-making efficiency and accuracy in the construction site acceptance process.

[0011] (2) This invention completely solves the fatal defects of existing general-purpose 3D rendering engines in the display of multiple labels, such as label stacking, crossing, or even being occluded by the model structure itself, by using a brand-new viewpoint adaptive routing and force-guided anti-occlusion label rendering algorithm. No matter what viewing angle the user is in, the dynamic labels can automatically find blank areas in the screen or view frustum for floating display by relying on the built-in spring force and repulsive force, providing an excellent visual interactive experience for large-scale engineering deviation verification with multiple concurrent points.

[0012] (3) In addition to displaying color changes at the point level, the dynamic chromatographic mapping and voxel warning mechanism of the present invention further provides regional macroscopic warnings for out-of-tolerance areas through density evaluation within spatial voxel blocks. This visual communication method, from point to volume and from micro to macro, abstracts complex engineering errors into intuitive thermal density warnings, which perfectly meets the requirements of computer-aided analysis and comprehensive evaluation under the framework of complex systems. Attached Figure Description

[0013] Figure 1 System overall architecture diagram; Figure 2 A schematic diagram illustrating the workflow of three-dimensional spatial deviation simulation calculation and chromatographic mapping; Figure 3 Logical evolution sequence diagram of viewpoint adaptive routing and force-oriented layout algorithm; Figure 4 A visual display and decision-making sequence diagram for the acceptance of multi-level complex building structures. Detailed Implementation

[0014] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following description is provided in conjunction with the appendix. Figure 1-4 The present invention will be further described in detail below with reference to the embodiments. It should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention. Unless otherwise specified, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0015] All mathematical formulas and the variables and functions contained in the formulas appearing in this manual have clear and public physical and computational meanings given in the context. To ensure the rigor and unambiguity of this manual, this manual follows the principle of one unique character corresponding to one unique physical parameter, and does not use the same character to represent different meanings in the same document.

[0016] Example 1 This embodiment focuses on detailing the core data flow architecture of a 3D model visualization comparison system for construction acceptance, as well as the underlying implementation of the quantitative calculation of 3D spatial deviation and the chromatogram out-of-tolerance mapping rendering mechanism. This system is designed specifically for digital twin systems of large and complex engineering systems and strictly follows the simulation and comparative analysis specifications of computer-aided design.

[0017] Referring to the overall system architecture, this comparative system is mounted on a high-performance edge computing server or cloud graphics workstation, and its first major core module is the measured data analysis and fusion module.

[0018] At the construction site, the engineering team typically uses high-precision sensors such as 3D laser scanners and UAV oblique photography to digitally reconstruct the physical entities of completed buildings (such as steel structure nodes and electromechanical integrated pipe corridors) to obtain 3D reconstructed point cloud data. Since this point cloud data is in a local instrument coordinate system, it cannot be directly compared with the initial design 3D model data built based on the global geodetic coordinate system or engineering coordinate system. The measured data analysis and fusion module first uses the Voxel Grid downsampling algorithm to downsample the raw point cloud at the level of hundreds of megabytes or even gigabytes, and uses statistical filtering (SOR) to remove free noise points caused by multipath effects or air dust while reducing data redundancy.

[0019] After preprocessing, the module uses a global coarse registration algorithm based on feature fast point histogram (FPFH) for initial alignment, and then uses an improved Iterative Closest Point (ICP) matching algorithm to achieve fine registration. By continuously minimizing the mean square error between the point cloud and the design model patch set, the geometric coordinates of the actual as-built point cloud are seamlessly unified into the isomorphic 3D space of the initial design 3D model, generating a 3D spatial data base that can be used for subsequent source-by-source comparison and fusion.

[0020] On this isomorphic three-dimensional spatial basis, the system flows to the second major core module: the three-dimensional spatial deviation simulation calculation module.

[0021] In order to quantify construction errors in a large and complex component topology network, it is necessary to calculate the distance between the measured completion point and the design expected surface. This module quickly performs the nearest neighbor geometric surface search by constructing a bounding box hierarchical structure (BVH tree) or a KD-Tree (K-dimensional tree) spatial data index structure.

[0022] For the above calculation process, this embodiment introduces Formula 1 for detailed explanation: Formula 1 is used to express the minimum Euclidean distance from a spatial point to the model: ; In Formula 1 above: The first point in the set of measured data points The coordinates of a measured data point (usually including the X, Y, and Z values ​​of three-dimensional Cartesian coordinates).

[0023] This represents the initial 3D design model that has been parsed and loaded into system memory.

[0024] Represents the initial design 3D model The first continuous mesh segmentation on the surface The coordinates of each grid point.

[0025] The standard Euclidean 2 norm operator.

[0026] This represents the point obtained through calculation. Compared with the initial design 3D model The minimum spatial distance function between them.

[0027] Through concurrent thread processing, the three-dimensional spatial deviation simulation calculation module calculates millions of... The corresponding absolute distance and the sign of the normal deviation (positive sign indicates convexity relative to the design surface, negative sign indicates concavity) are encapsulated in the deviation data set in the memory pool.

[0028] After calculating massive amounts of spatial deviation data, directly sending the reports to the on-site engineers would be insufficient to effectively guide acceptance due to the dimensional overload of the data. Therefore, the system enters its third core module: dynamic chromatographic mapping and voxel warning module.

[0029] This module abandons the approach of simply labeling or listing numbers. Instead, it uses fragment shaders based on a programmable rendering pipeline (WebGL or related underlying graphics APIs) to translate the deviation of each measurement point or the as-built surface after mesh reconstruction into visual color values.

[0030] To address this, Formula 2 is introduced to explain the logic of the chromatographic warning mapping function: ; In Formula 2 above: This represents the vector of rendered color values ​​output to the final frame buffer in the graphics card rendering pipeline, and is represented in (R, G, B, Alpha) format.

[0031] This refers to the pre-set tolerance threshold parameters based on relevant national or industry acceptance standards for building construction, used to define the acceptable range for errors within a certain number of millimeters (mm).

[0032] This represents a predefined multi-segment chromatographic mapping function, when the deviation... When it is close to 0, it is mapped to a cool color (such as green or blue) that represents safety. when Gradually increase until approaching and exceeding At that time, the color transitions smoothly from warm tones and warning colors (such as yellow, orange, and even bright pure red) in a non-linear exponential or linear manner. For redundant values ​​that exceed the threshold, force them to be highlighted in red.

[0033] Through this mapping process, the originally monotonous surface of the 3D as-built model is covered with a gorgeous 3D color spectrum containing precise error magnitude attributes. When engineers zoom in and out of the model at will, large red patches can intuitively highlight areas where the construction deviates significantly from the design, just like "lesions," without having to search for problems in the vast sea of ​​data in 2D reports.

[0034] Example 2 Although overall deviation visualization was achieved through chromatographic mapping in Example 1, key out-of-tolerance features (such as misalignment of embedded parts, displacement of the center of bolt holes in steel structures, etc.) must be accurately reported through "labels" containing text values ​​and cause prompts. To address this requirement, this example provides a very detailed explanation of the fourth core module of the system, which is highly innovative: the viewpoint adaptive routing and label rendering module.

[0035] In actual 3D software operation, since the user's viewport (Camera) can be arbitrarily traversed (Pan, Orbit, Zoom), if a 3D text box with world coordinate system or screen coordinate system size is directly generated on a 3D point (anchor point) in space that produces serious errors, when a large number of deviations are concentrated in a narrow area of ​​a certain electromechanical pipeline, the text boxes will inevitably overlap and intertwine to form unreadable chaotic blocks. Additionally, labels located on the back or deep within the model, if not properly masked, may interfere with observation.

[0036] To address this technical challenge, this module does not employ a rigid screen floating mechanism. Instead, it uses a dynamic simulation system based on wire connection and force-directed repulsion to achieve dynamic rendering. Each deviation label can be considered a point mass with virtual mass. They are subject to the spring tension generated by their own anchor point, the repulsive force between similar labels, and the three-dimensional collision blocking force generated by the bounding box of obstacles in the viewpoint.

[0037] To calculate the trajectory that avoids obstruction by the model body and achieves maximum visual clarity, this module needs to perform line-of-sight analysis and penalty mechanism calculation. We evaluate whether model penetration occurs by emitting false rays from the center of the label to the viewpoint of the observation camera.

[0038] At this point, we introduce Formula 3 to calculate the viewpoint occlusion cost function: ; In formula 3 above: This indicates the first label anchor point connected to the label screen hanging position. A set of geometric line segments for candidate lead paths.

[0039] Indicates the measurement of the first The occlusion cost function is the degree to which a candidate lead path is occluded by other structures in the scene within the 3D view frustum projection.

[0040] This represents the total number of 3D obstacles (model components) within the current field of view, as filtered by the camera's Frustum Culling algorithm.

[0041] Indicates the first The geometric parameters of the three-dimensional axial bounding box (AABB bounding box) of the obstacle being screened.

[0042] This represents a ray-based intersection determination function, when candidate lead paths... Projected rays and obstacle enclosure When geometric overlap occurs, the function outputs 1; Conversely, if the line of sight is clear and unobstructed, the output is 0.

[0043] Indicates that for the first The occlusion weight parameter assigned to each obstacle is dynamically adjusted according to the proportion of the projected area occupied by the obstacle in the screen space.

[0044] Besides avoiding interference with the model, the optimal lead wire should balance the simplicity of its length and the smoothness of the broken line. Therefore, a joint optimization objective is constructed, and Equation 4 is introduced to expand the objective function for lead wire path optimization: ; In Formula 4: This represents the optimal lead path extracted and ultimately selected as the rendering result after minimizing the objective function.

[0045] This represents a mathematical function used to calculate the total curve length of candidate leads.

[0046] This represents the length weighting coefficient used to control the ratio of path length to total cost.

[0047] This represents the occlusion cost weighting coefficient corresponding to the calculation result of Formula 3 above.

[0048] It represents a function that measures the overall unevenness of the lead wire's bends, and is obtained by accumulating the deflection angles at each turning point on the lead wire.

[0049] This represents the penalty weighting coefficient for the lead bending angle, used to force the lead to tend towards a straight line or a smooth curve with a single bend.

[0050] The static lead-finding mechanism described above ensures that the lead does not clip through the mold. In order to solve the problem of multiple labels overlapping, this embodiment embeds differential equation integration operations between consecutive rendering frames (frame-by-frame) so that the labels automatically disperse like mutually exclusive particles in the physical universe.

[0051] The advancement of this layout relies on the following dynamic definition system, specifically based on the kinematic formulas of spring force, repulsive force, and force balance.

[0052] First, we introduce Formula 5 to define the lead wire tension (spring force): ; In formula 5 above: This function represents the spring force calculation function that forces the label to return to a reasonable range of its corresponding deviation anchor point and prevents the label from drifting infinitely in the viewport. The output is a three-dimensional force vector.

[0053] This represents the current center point coordinates of the deviation label in three-dimensional space.

[0054] The coordinates of the anchor point fixed on the 3D model represent the actual physical location where the deviation was detected in space.

[0055] This indicates the spring constant used to adjust the stiffness of the lead wire. The larger the value, the closer the label is to the anchor point.

[0056] This represents the ideal distance threshold parameter between tags and anchors that aligns with optimal human cognitive reading habits.

[0057] To reduce crowding, Formula 6 is then introduced to define the repulsive force between labels: ; In formula 6 above: This represents the result of a repulsive force function calculated based on a form similar to Coulomb's law between any two different biased labels with overlapping risk. This force vector causes the labels to escape from each other.

[0058] This indicates that the force analysis is in progress. The three-dimensional center position of each deviation label.

[0059] Indicates that the current enclosure is in the first... The first tag retrieval radius contains interference sources. The three-dimensional center position of each deviation label.

[0060] This represents the repulsion coefficient, which is used to control the strength of the repulsion force.

[0061] Finally, the dynamic adjustment of the label position is performed by explicit Euler integrals. Formulas 7 and 8 are introduced to explain the rules for force equilibrium iteration and position update within the system: (Formula 7); (Formula 8); In formulas 7 and 8 above: This indicates that after a round of physical collision detection and mechanical synthesis calculations, the system assigns the first... The new frame velocity vector of each label.

[0062] This represents the velocity damping attenuation coefficient (ranging from 0 to 1) used to suppress continuous oscillations in the system and promote its eventual convergence to rest.

[0063] Indicates the first The old velocity vectors that are retained from the previous rendered frame for each label.

[0064] It represents a scalar value indicating the time step between two rendering frame refreshes (for example, for a 60fps refresh rate, it is usually about 0.0167 seconds).

[0065] Substitute the result of the calculation in Formula 5 above into the... The specific resultant force value of the spring force obtained after the specific variables of each label.

[0066] Substitute the result of the aforementioned formula 6 into the corresponding... With the The resultant force value obtained by calculating the repulsive force parameters of each label.

[0067] This represents a virtual constant scalar value of mass pre-assigned to the deviation label in a virtual simulation digital system.

[0068] The representation is the first term obtained through the position derivative update operation. The core anchor coordinates of the label that will be drawn in the current rendering frame.

[0069] Representing the first The tag is the residual position vector from the previous frame.

[0070] Through the closed-loop calculations of the aforementioned force-guided dynamic algorithm and line-of-sight avoidance algorithm, as construction managers rotate their viewing angle in front of the browser or desktop workstation, the previously chaotic numerous out-of-tolerance annotation boxes on the screen will automatically, like living things, gracefully slide and automatically expand and contract according to the available space. Without interfering with each other's display areas, each bending leader line can smoothly point to the out-of-tolerance point. Due to the damping coefficient... The existence of this feature means that the motion will quickly return to a stable lock after the viewpoint stops rotating. This viewpoint-adaptive anti-occlusion mechanism maximizes the translation efficiency of three-dimensional deviation positioning in a two-dimensional plane.

[0071] Example 3 Based on Embodiment 1 and Embodiment 2, considering the complexity of the on-site scene of ultra-large-scale civil construction projects, the measured point cloud of tens of millions or even hundreds of millions of points may have hundreds of thousands of tiny errors at the same time. If all errors are generated and displayed with labels, even the most powerful anti-occlusion algorithm will lead to visual disaster and complete collapse of rendering computing power. Therefore, this embodiment proposes a deep evolution scheme for the hierarchical display of data in ultra-large-scale scenes, which includes spatial voxel-based dynamic mesh division and thermal density warning mechanism.

[0072] The system in this embodiment adds a spatial structure preprocessing and distribution layer: first, the three-dimensional global bounding box size of the entire project is extracted, and then the entire three-dimensional architectural space is regularly divided into three-dimensional voxel grids with different side lengths according to a specific octree spatial subdivision hierarchy strategy. The system does not assign computational load to every tiny qualified deviation, but introduces a calculation mode based on micro-integration to macro-density.

[0073] Formula 9 is introduced to illustrate the statistical model of the out-of-range cluster density of spatial voxel blocks: ; In Formula 9 above: This represents the deviation density value obtained through statistical calculation to measure the severity of local construction quality loss (this value is represented as a numerical matrix attached to the corresponding spatial index voxel attribute).

[0074] It represents a cubic voxel block that occupies a specific region of coordinate boundaries in three-dimensional space after being partitioned by a meshing algorithm.

[0075] This represents the physical spatial volume scalar parameter inherent in the metric category of the aforementioned three-dimensional spatial cube voxel block.

[0076] This represents the standard step function calculation logic. When the internal operation value passed to this function (i.e., the measured distance minus the tolerance) is... When the value is strictly greater than 0 (indicating a substantial deviation at this measured point), the step function output value is 1; Conversely, if the measurement point is within the allowable range, the step function ignores negative numbers and forces an output of 0. This is achieved by applying the entire... All measured points contained within a voxel By iterating and accumulating, the total number of actual violations is obtained and divided by the volume.

[0077] Using the voxel density data stream provided by Formula 9, the system utilizes a graphics rendering interface to establish a volume rendering warning mechanism similar to atmospheric rendering or medical MRI volume rendering. When users view the overall BIM model of the entire large steel structure stadium or super factory, there is no need to identify the color spectrum of individual components at a distance from the camera. In areas with higher voxel density, the system automatically emits high-frequency, low-transparency, and breathing-like red pulse highlights (shader material rendering effects) within its voxel bounding box. This is a "thermal cloud" warning effect: it directly and quickly tells the project manager that "the overall error density of the northeast corner roof load-bearing area is extremely dense, and there is a systemic risk of structural misalignment."

[0078] When the administrator clicks on the area where the red warning cloud is located, the camera viewport will quickly zoom in. When the camera's frustum distance shortens beyond a specific Level of Detail (LOD) threshold, the system automatically hides the macroscopic red pulse warning voxel in that area and smoothly loads the fine-level surface chromatogram texture from Example 1. Subsequently, the powerful force-guided layout module from Example 2 is activated, automatically growing and scattering clear and neatly arranged detailed text deviation labels from the severely out-of-tolerance nodes. The text content can be linked in real time to display the specific parameters of the construction, such as "Embedded component 13B: Z-axis sinking 53mm, exceeding the tolerance specification value of 30mm".

[0079] By employing a multi-level, interconnected 3D presentation approach—from macroscopic thermal density alarms and mesoscopic surface continuous chromatographic mapping to microscopic intelligent anti-obstruction tag hovering—rigid, isolated data is transformed into a highly vivid and interactive audiovisual experience system. This not only allows decision-makers to quickly identify macroscopic quality defects but also ensures that frontline construction technicians can accurately locate 3D components requiring rework and repair thanks to excellent view resolution. This establishes a closed-loop data transmission system from quality inspection to corrective action, maximizing the powerful commercial and industrial application value of this system-classified 3D visualization comparison in the real-world digital twin industry.

[0080] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it; Those skilled in the art should understand that: It is still possible to modify the technical solutions described in the foregoing embodiments, or to make equivalent substitutions for some of the technical features; These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A 3D model visualization and comparison system for construction acceptance, characterized in that, The system is based on computer-aided design and digital twin architecture, including a measured data analysis and fusion module, a three-dimensional spatial deviation simulation calculation module, a dynamic chromatographic mapping and voxel warning module, and a viewpoint adaptive routing and label rendering module. The measured data parsing and fusion module is used to acquire the three-dimensional reconstruction point cloud data of the construction site and the initial design three-dimensional model data. It uses a three-dimensional registration algorithm to register the three-dimensional reconstruction point cloud data to the unified global coordinate system where the initial design three-dimensional model data is located, so as to realize the fusion and alignment of multi-source heterogeneous models in three-dimensional space. The three-dimensional spatial deviation simulation calculation module is communicatively connected to the measured data parsing and fusion module. It is used to calculate the minimum Euclidean distance of each aligned measured data point relative to the surface of the initial design three-dimensional model using the nearest neighbor search space algorithm, extract all spatial deviation data, and establish a deviation dataset containing spatial coordinates and deviation amounts. The dynamic chromatographic mapping and voxel warning module is communicatively connected to the three-dimensional spatial deviation simulation calculation module. It is used to map the deviation amount in the deviation dataset into RGB rendering color values ​​in real time based on the preset out-of-tolerance threshold and chromatographic mapping function, and seamlessly attach and cover the RGB rendering color values ​​to the surface of the measured data points to form a three-dimensional chromatogram. Simultaneously, the global three-dimensional space is divided into multiple voxel blocks, and the out-of-tolerance density value within each voxel block is counted. When the out-of-tolerance density value exceeds the safety threshold, a pulsed highlight warning is triggered in the local space. The viewpoint adaptive routing and label rendering module is communicatively connected to the dynamic color mapping and voxel warning module, and is used to capture the observation angle of the current viewport camera in real time during digital twin 3D scene roaming. For anchor points with significant deviations in the deviation dataset, force-directed layout and view frustum anti-occlusion target optimization algorithms are used to calculate non-overlapping guide paths and label floating coordinates that avoid obstruction by the model body, and dynamic deviation labels with guides are rendered in real time in the 3D scene.

2. The 3D model visualization and comparison system for construction acceptance as described in claim 1, characterized in that, The measured data parsing and fusion module preprocesses the 3D reconstructed point cloud data, including: downsampling the original point cloud using a voxel mesh downsampling algorithm, and removing detached noise points using statistical filtering; After preprocessing, a global coarse registration algorithm based on feature fast point histogram is used for initial alignment, and then an iterative nearest point matching algorithm is used to achieve fine registration. By continuously minimizing the mean square error between the point cloud and the design model patch set, the geometric coordinates of the actual as-built point cloud are unified into the isomorphic 3D space of the initial design 3D model.

3. The three-dimensional model visualization and comparison system for construction acceptance as described in claim 1, characterized in that, The three-dimensional spatial deviation simulation calculation module performs nearest neighbor geometric surface search by constructing a spatial data index structure; For each measured data point, calculate its minimum Euclidean distance to the surface of the initial design 3D model using the following formula: ； in, The first point in the set of measured data points The coordinates of each measured data point include the X, Y, and Z values ​​of three-dimensional Cartesian coordinates; Represents the initial design 3D model; Represents the initial design 3D model The first continuous mesh segmentation on the surface Coordinates of grid points; Operators representing Euclidean norm 2; Point Compared with the initial design 3D model The minimum spatial distance function between them; The three-dimensional spatial deviation simulation calculation module processes data through concurrent threads, calculates the absolute value of the distance and the sign of the normal deviation corresponding to each measured data point, where a positive sign indicates a convexity relative to the design surface and a negative sign indicates a concavity, and encapsulates the data into a deviation dataset.

4. The 3D model visualization and comparison system for construction acceptance as described in claim 1, characterized in that, In the dynamic chromatographic mapping and voxel warning module, for the deviation of each measured data point, a fragment shader based on the programmable rendering pipeline is used to translate the deviation into a rendered color value through a predefined multi-segment chromatographic mapping function. When the deviation is close to 0, it is mapped to a cool color that represents safety. When the deviation gradually increases until it approaches and exceeds the preset out-of-tolerance threshold, it smoothly transitions to warm colors and warning colors in a non-linear exponential manner. For redundant values ​​that exceed the threshold, force them to be highlighted in red.

5. The three-dimensional model visualization and comparison system for construction acceptance as described in claim 1, characterized in that, In the dynamic chromatographic mapping and voxel warning module, the entire three-dimensional architectural space is divided into three-dimensional voxel blocks of different side lengths according to the octree spatial subdivision hierarchy strategy. The out-of-tolerance density value within each voxel block is calculated using the following formula: ； in, This represents the density value of out-of-tolerance deviation; This represents a cubic voxel block that occupies a specific region of coordinate boundaries in three-dimensional space after being meshed. The physical space volume scalar parameter representing the cube voxel block; Represents the step function, when Output 1 if the value is strictly greater than 0, otherwise output 0; Represents the measured data points Compared with the initial design 3D model The minimum spatial distance between them; This represents the preset out-of-tolerance threshold parameter; When the out-of-tolerance density value exceeds the safety threshold, the system renders a breathing-like, pulsating red pulse highlight warning effect inside the corresponding voxel block bounding box.

6. The three-dimensional model visualization and comparison system for construction acceptance according to claim 1, characterized in that, In the viewpoint adaptive lead path routing and label rendering module, the degree of occlusion of candidate lead paths is evaluated by constructing a viewpoint occlusion cost function, specifically calculated according to the following formula: ； in, This indicates the first label anchor point connected to the label screen hanging position. A set of geometric line segments for candidate lead paths; Indicates the first The occlusion cost function for the degree to which a candidate lead path is occluded by other structures in the scene within the 3D view frustum projection; This represents the total number of 3D obstacles within the current field of view, as filtered by the camera clipping space culling algorithm. Indicates the first Geometric parameters of the three-dimensional axial bounding box of the obstacle being screened; This represents a ray-based intersection determination function, when candidate lead paths... Projected rays and obstacle enclosure Output 1 when geometric intersection occurs, otherwise output 0; Indicates that for the first The occlusion weight parameter assigned to each obstacle is dynamically adjusted according to the proportion of the projected area occupied by the obstacle in the screen space.

7. The three-dimensional model visualization and comparison system for construction acceptance as described in claim 1, characterized in that... ； in, This represents the optimal lead path selected as the rendering result after minimizing the objective function. This represents a mathematical function used to calculate the total curve length of candidate leads; This represents the length weighting coefficient used to control the ratio of path length to total cost. This represents the occlusion cost function. The occlusion cost weighting coefficient; The function representing the overall unevenness of the lead wire's bends is obtained by summing the deflection angles at each turning point on the lead wire; This represents the penalty weighting coefficient for the lead bending angle, used to force the lead to tend towards a straight shape.

8. The 3D model visualization and comparison system for construction acceptance as described in claim 1, characterized in that, In the viewpoint adaptive routing and label rendering module, each deviation label is treated as a point mass with virtual mass and dynamically laid out under the action of spring force and repulsive force. The formula for calculating spring force is: ； in, This represents the spring force that forces the label to revert to a reasonable range of its corresponding deviation anchor point, and the output is a three-dimensional force vector. This represents the current center point coordinates of the deviation label in three-dimensional space. The coordinates of the anchor point fixed on the 3D model represent the actual physical location of the deviation. This indicates the spring constant used to adjust the stiffness of the lead wire; This represents the ideal distance threshold parameter between the label and the anchor point. The formula for calculating the repulsive force between tags is: ； in, This represents the repulsive force between two different deviation labels, established based on the form of Coulomb's law. Indicates the first The three-dimensional center position of each deviation label; Indicates the first The three-dimensional center position of each deviation label; This represents the repulsion coefficient, which is used to control the strength of the repulsion force.

9. The three-dimensional model visualization and comparison system for construction acceptance as described in claim 1, characterized in that, The label position is adjusted dynamically by explicit Euler integration, and the velocity update formula is: ； The position update formula is: ； in, Indicates the first The velocity vector of the new frame for each label; This represents the velocity damping attenuation coefficient used to suppress continuous oscillations in the system and promote convergence to rest. Indicates the first The old velocity vector of each label in the previous rendered frame; A scalar representing the time step between two rendering frame refreshes; For the first The resultant force value of the spring force on each label; For the first With the The resultant force of the repulsive force between the labels; This represents a virtual, constant scalar value of mass pre-assigned to the deviation label; Representing the The anchor coordinates of the label that will be drawn in the current rendering frame; Representing the The residual position vector of each label in the previous frame.

10. The three-dimensional model visualization and comparison system for construction acceptance according to claim 5, characterized in that, The system has a multi-level linked 3D presentation mechanism: When the user views the overall model from a distance, a macroscopic red pulse highlight warning driven by voxel density is displayed. When the user zooms in through interactive operations, and the distance between the camera's truncated cone and the camera's truncated cone exceeds the spatial level threshold, the system automatically hides the macroscopic red pulse warning voxel and smoothly transitions to load a fine-level surface chromatogram texture. At the same time, the force-guided layout module is activated, automatically generating and distributing neatly arranged detailed text deviation label leads from the out-of-tolerance nodes.

Images